Normalization: A Tool for Controlling Concentration Effects in Fuzzy Set Aggregation

This paper addresses the concentration effect, a common issue in fuzzy set aggregation, particularly when using operators like the minimum, which can overemphasize lower membership degrees. We propose normalization as a simple yet effective technique to mitigate this problem, introducing two distinct approaches: pre-aggregation normalization, where individual fuzzy sets are normalized before aggregation, and postaggregation normalization, where the aggregated set is normalized. We analyze the impact of these methods on key aggregation properties, such as idempotency and monotonicity, revealing a trade-off between property preservation and concentration reduction. Using min-max normalization as a case study, we demonstrate through experimental results that both approaches, especially post-aggregation normalization, effectively counteract concentration effects, often matching or outperforming more complex techniques while maintaining computational efficiency. The paper provides a detailed algorithmic implementation of both approaches, enabling easy integration into existing fuzzy logic systems. The simplicity, adaptability, and effectiveness of normalization make it a valuable tool for enhancing the accuracy, interpretability, and robustness of fuzzy set aggregation in diverse applications, including multi-criteria decision making, pattern recognition, fuzzy control, and image processing.